DFB1 Support thread

[exxos]

Sorry if this is dumb.. but I've not been following much. But can't alt-ram just be disabled on boot like on the ST ? Then let HD11 load as normal, then enable alt-ram and install the FRB manually with alt-ram ?

[Badwolf]

Sorry if this is dumb.. but I've not been following much. But can't alt-ram just be disabled on boot like on the ST ? Then let HD11 load as normal, then enable alt-ram and install the FRB manually with alt-ram ?

The problem’s been solved, so this is moot, but ultimately this doesn’t work as the order things are done is almost irrelevant.

Basically HDDriver tries to transfer (much) more than 64kB at a time when using SCSI on a Falcon. Even if it’s using an FRB (which is 64kB). An understandable mistake (SCSI on a stock TT or Falcon doesn’t need an FRB), but a fairly big mistake nonetheless. If you are going to use the FRB, you need to be aware of its size too.

We knew something was wrong with HDDriver but wasn’t sure what. Uwe suggested some diagnostics which no one who was affected managed to follow up on (likely as none of us are SCSI protocol experts and so probably couldn’t even understand what we would be seeing). He seems reticent to explain his program’s logic, which hampers diagnosis, but I suppose that is the purview of a propriety software vendor.

Anyway, Frank provided the key bit of information via ingenious trial and error that pointed to a buffer overflow: he could load or run programs under 64kB, but not larger.

To test this theory, Frank and I allocated 128kB for an FRB. Sure enough, files or programs up to 128kB now worked.

After that Uwe identified the problem (he wasn’t aware of the Falcon’s hardware restrictions) and has proposed a fix. There was still just enough time for him to insult his customers a couple more times, but I hope now he’ll provide a solution.

Now, this might get me into trouble but I would genuinely like to know if he is courteous and helpful in German, or whether it’s only with certain people? I’ve been admonished in the past for perhaps being uncharitable in my assessment of my customer service experience at HDDriver.net. Now, hands up, my German is extremely basic and I wouldn’t know if I ought to be offended when conversing in it, but In English I can. And I am.

BW

[Rustynutt]

Think you're fine BW.
You just got to experience what quite a few others have asking questions :)

And they were good ones, which picked close to HDDrivers actual operation. There were a couple statements he made I was a bit taken back by.

Only thing I questioned is you state AHDI works. Thought a few pages back the thread a user said AHDI did not work? I know when I posted the legacy driver archive you stated it's already been determined AHDI does not work.

So you set Frank up with the TOS you integrated MAPROM (or fast RAM.prg) into a TOS binary, and he burned an EPROM, or flashed the DFB1?
And you also provided a resized FRB?

Pavel of Centuriontech built me an EPROM flash that sets in the ROM socket.
I can burn TOS binaries now without going through the expensive EPROMS :)

The 5 UV erasable ones I'd bought from China last year, only 2 worked, and both are hit and miss if they erase.

I'd like to test your modification. If allowed, could I also request you edit the binary to set the Falcon control register bit setting for the Blitter to 1/2 (8MHz)?

Think you're able to correct the CRC AIRC. I'm don't have an environment setup still to compile source, like Steve Usher provides here.


If so, you can send the binary via messenger, or email here.

Isn't this fun? :)
Thanks!

[exxos]

Now, this might get me into trouble but I would genuinely like to know if he is courteous and helpful in German, or whether it’s only with certain people? I’ve been admonished in the past for perhaps being uncharitable in my assessment of my customer service experience at HDDriver.net. Now, hands up, my German is extremely basic and I wouldn’t know if I ought to be offended when conversing in it, but In English I can. And I am.

Everyone who deals with him says the same from what I have heard. He seem polite on a first email and becomes irritated if you email him again. I asked for help when we developed Trudie as one version of his driver worked and then something changed and it didn't work anymore. I asked what he changed to help diagnose and got nothing. There's small bugs in his driver i found but can't be bothered anymore.

Techie Alison once asked for a free copy of the driver when she was developing a hard drive and said she would promote his driver when the final product was done, he basically told her to sod off. It's not like we are nobodies in the Atari world. People who want to help develop my hardware I give them stuff mostly for free. But parts and assembly is expensive so its not always possible. Sadly the only other person doing drivers is even more difficult to deal with. Its why I said a while back we really need a open source driver as like you found out, its just hampering efforts when we can't get decent help.

Word on the street is Germans can be abrasive. But its not just that. Good he's still updating the driver, but always frowned upon as nobody charges for software these days. It be like me charging for GB7 for all my work. Or you charging to update your software. If everyone charged there wouldn't be anyone using Atari's anymore. The community isn't that large anymore as it is. I generally always try and help people as much as I can. The whole thing gets very tiresome unfortunately. :(

[czietz]

I still find it deeply unfair to attack someone in a forum where they won't present their side of the story. I still think any grudges, misunderstandings and such are best settled in private. But I don’t mind at all publicly standing by what I wrote you some weeks ago: Personally, I have nothing but praise about Uwe's support for HDDRIVER, whenever I needed it: Fast and professional replies and the ability to discuss problems on a very detailed technical level.

[exxos]

I still find it deeply unfair to attack someone in a forum where they won't present their side of the story. I still think any grudges, misunderstandings and such are best settled in private. But I don’t mind at all publicly standing by what I wrote you some weeks ago: Personally, I have nothing but praise about Uwe's support for HDDRIVER, whenever I needed it:

Nobody is attacking him. And certainly nobody is commenting on the quality of his software. It is why @Badwolf Mentioned if this was down to language barrier or something else.

whenever I needed it: Fast and professional replies and the ability to discuss problems on a very detailed technical level.

I have never had that and I don't know anyone else who I have known who has dealt with him. I just gave up in the end asking for help. Again also why @Badwolf Asked if he only help certain people or just unhelpful in general. I don't see any harm in just asking the question as I would really like to know the answer myself as well.

Indeed if anyone thinks I'm being unfair or anyone else is being unfair then I will gladly remove the comments.

[czietz]

Meanwhile, HDDRIVER 11.14 was released today with the fix that Frank had already successfully tested: https://www.hddriver.net/en/news.html.

[Rustynutt]

The problem’s been solved, so this is moot.

Anyway, Frank provided the key bit of information via ingenious trial and error that pointed to a buffer overflow: he could load or run programs under 64kB, but not larger.

To test this theory, Frank and I allocated 128kB for an FRB. Sure enough, files or program.

BW

Ok, currently testing the patch in V11.14 of HDD on AB, "stock" bus.

As Uwe pointed out for the previous version, this slows transfers considerably.

So far, 1GB file transfers (from 1k files to those 10mb in size) to SCSI CF results appear positive. However, I usually test over time, constantly running tree check and correct to ensure no errors appear.

Back to the DFB1. :)
So you've (well, Frank) been able to set the FRB at 128K, and successfully use the earlier version of HDDriver?

I'm all for following standards, but in the case of Falcon and alt RAM, it seems a standard is yet being defined.

Using the new HD Driver patch results in considerable SCSI transfer speed reduction.

If increasing the FRB to 128K works under Frank's testing of file sizes, wouldn't it make sense to just set an FRB size to the maximum block sizes HDDriver uses for file transfers and just be done with it?

I'm not good at this kind of math. In practice, the Falcon SCSI can reach ~1.6mb/s transfer rate. I know ST RAM speed is much faster on the DFB1 than the Afterburner, but the AB can still manage double that rate of SCSI.

It seems opening up the FRB to a maximum size for DFB1 TOS (and hopefully those that can flash their own ROMS) would be a far better solution than offered with the autofolder patch, SCSI standards be damned?

After all, all we are look for is to have DMA passed to ST RAM before writing to the bus.

Make sense?

[Rustynutt]

Ok, few hours testing.
2 things to consider.
1: This is an Afterburner, not DFB1.
2: The clock patch does not have the resistor on the DMA clock line as Exxos recommends. Was saving that mod until I can see on a scope what changes between a 16MHZ and 23MHz bus clock on the SDMA signal.

Using the "new" autofolder patch, after transferring ~1GB files from IDE to SCSI CF, Falcon bus 32/16, I get file check errors.

Same test, bus 30/15, I get no errors, even without the autofolder patch.

So, at this point, going to say until my scope arrives, I can't really say anything defining.

[Badwolf]

I still find it deeply unfair to attack someone in a forum where they won't present their side of the story.

I don't want to drag this out and I certainly don't want to perform a hatchet job on the guy, but all sides of the story are on his forum. The first thing he does is tell me -- a free hardware designer -- off for not testing my board (that someone else has built) with SCSI as if my (correct) identification of his driver as the problem never happened and it's somehow my fault!

I know this was done with a smiley, but that's simply rude. Things don't get any better later on.

Personally, I have nothing but praise about Uwe's support for HDDRIVER, whenever I needed it: Fast and professional replies and the ability to discuss problems on a very detailed technical level.

I wonder if this is one of those things where you have to have reached a particular threshold in your understanding of the issue before you see this? All I ever seem to get is referred to standards or (unnamed) books. I always prefer to see 'cock up rather than conspiracy', though, so perhaps it's that. Or perhaps those exchanges were in his native language & that's a factor?

Anyway, I do agree with you we ought not dwell on an individual here. Firstly it's off topic but secondly it's is very unedifying, and I apologise for that. I needed to let off a bit of steam without burning all bridges. Don't feel the need to answer my question above if you're prefer not to.

Cheers,

BW